Technology has always changed employment in two directions at once. It can reduce demand for established jobs, but it can also create entirely new professions built around skills that did not previously exist. As artificial intelligence spreads through workplaces in 2026, the central question is becoming more urgent: will AI create valuable new careers, and who will be most likely to access them?
A new study led by labour economist David Autor examines decades of employment data in the United States to understand how new technology-driven jobs have emerged in the past. Its findings suggest that newly created work has historically benefited younger, highly educated workers in urban areas more than other groups. The research also points to a broader lesson for the AI era: new jobs do not simply appear because a technology exists; they are often shaped by investment, demand and policy decisions.
New Technology Does Not Benefit All Workers Equally
The study focuses on what researchers describe as “new work”: occupations and specialised roles that emerge as technology, industry and demand change. An earlier study by the same researchers found that around six in ten jobs in the United States between 1940 and 2018 were in fields that had developed broadly after 1940.
The latest research goes further by asking who actually filled those new roles.
Using United States Census Bureau data from 1940 to 1950 and American Community Survey records from 2011 to 2023, the researchers compared people working in newly created occupational specialties with those in more established jobs.
The results show a clear pattern. New forms of work were more concentrated in cities, more likely to be filled by people under the age of 30, and more accessible to workers with university education.
In 1950, around 7 percent of employees were working in job categories that had emerged since 1930. In the period from 2011 to 2023, approximately 18 percent of workers were employed in forms of work introduced since 1970.
The research also found that workers who entered new occupations early were more likely to remain connected to new fields later. People employed in newly emerging work in 1940 were 2.5 times more likely to be working in new occupational areas in 1950 than workers in the wider population. University graduates were also 2.9 percentage points more likely than high school graduates to work in newly developed roles.
Why New Jobs Often Pay More at First
One reason new technology-driven jobs can offer higher wages is that they depend on knowledge that is still rare.
When a new technology appears, relatively few people initially know how to use it, build it, manage it or apply it to a business problem. That scarcity can make skilled workers more valuable and lead to better salaries.
But this advantage does not necessarily last forever.
As a skill becomes more common, companies can find more workers capable of doing the same work. Training becomes easier, tools become simpler, and some parts of the job may eventually be automated. The wage advantage connected to early expertise can begin to decline.
History offers many examples. Driving a car was once a specialised capability. Using word-processing software was also considered an important office skill in the early computer era. Today, both are widely understood and no longer create the same professional advantage.
That pattern creates an important question for artificial intelligence. Skills such as prompting, using generative AI tools or automating basic office tasks may be valuable today, but some could quickly become standard workplace expectations rather than specialised careers.
Investment Helps Create New Kinds of Work
One of the most important findings of the research is that technology-driven jobs are not created by invention alone. Large-scale demand can play a decisive role.
The study examined employment patterns during the 1940s, when the United States government supported major expansions in research, manufacturing and industrial production during World War II. Counties that received new factories and industrial investment saw higher levels of newly created work.
According to the findings, between 85 and 90 percent of new work created from 1940 to 1950 was driven by technology. But that technology was closely connected to large public and private investment, which created demand for engineers, specialists, technicians and other new forms of expertise.
This historical lesson matters in the AI debate. Innovation may begin with researchers and technology companies, but good jobs are more likely to emerge when governments and industries invest in applying technology to real economic needs.
AI could automate tasks in existing roles without producing many new opportunities. Alternatively, it could be directed toward expanding industries, improving productivity and creating new areas of specialised employment.
The difference may depend on where investment goes.
What the Study Means for Artificial Intelligence Jobs
The arrival of artificial intelligence has created concern that businesses may use the technology mainly to reduce staffing costs. Tools capable of writing, analysing data, assisting customers or supporting administrative work may affect tasks currently performed by millions of employees.
However, replacing tasks is not always the same as eliminating entire jobs. Many occupations involve a combination of responsibilities, judgment, communication and specialised knowledge. AI may remove some activities while increasing demand for others.
The more difficult question is whether AI will generate enough high-quality new work to offset disruption in existing roles.
The study cannot predict exactly what AI jobs will emerge or who will receive them. But history suggests that new opportunities may initially favour young, educated workers who have access to technical skills, strong professional networks and major urban job markets.
That creates a challenge for policymakers and businesses. If AI-related job creation follows previous technology cycles, workers without university degrees, older employees and people outside major cities may find it harder to benefit unless training and investment are designed more broadly.
Healthcare Could Show Two Different Futures for AI
Healthcare offers one example of how AI could be deployed in very different ways.
One approach would be to use AI mainly to reduce the number of workers needed for existing tasks. Another approach would be to use the technology to help medical professionals, technicians and support workers perform more advanced responsibilities, improve access to care and raise productivity.
The second path could create new types of jobs and allow workers with different levels of expertise to contribute more effectively. But this outcome is not automatic. It depends on how hospitals, companies and governments choose to invest in the technology.
Because public spending represents a large share of healthcare funding in the United States, government decisions could influence whether AI becomes primarily a cost-cutting tool or a platform for expanding skilled work.
Will AI Repeat Technology’s Job-Creation Pattern?
The history of technology shows that new industries can produce new careers, higher wages and specialised expertise. But it also shows that the benefits are unevenly distributed and often temporary.
AI may create opportunities for new kinds of analysts, healthcare specialists, engineers, safety experts, educators and creative professionals. It may also make some existing tasks less valuable or easier to automate.
The main lesson from the study is that the future of AI employment is not predetermined. New jobs will depend on how the technology is implemented, where investment is directed, and whether workers are given opportunities to develop the skills needed for emerging fields.
For young and highly educated workers, AI could open the door to valuable careers that do not yet exist. For the wider workforce, the challenge will be ensuring that the benefits of new technology are not limited to those already best positioned to take advantage of it.